Berkshire Hathaway underperformed the S&P 500 by more than 30-percentage points over the last year! Berkshire's only annual performance that was worse was in 1999 -- during the Internet mania. Today, though, Berkshire owns Apple and Google, unlike in 1999 when it didn't own any tech stocks. So, what else could explain the company's declining performance? More than a decade ago, researchers at AQR ran a series of regression studies across 30 years of Berkshire's public stock investments to discover which factors drove Buffett's outstanding investment results. They discovered -- to no one's surprise -- that Buffett buys ultra-high quality, large-cap, low-volatility stocks that are extremely cheap. But that's not all they discovered. They also disaggregated Buffett’s portfolio into two distinct sleeves and then ran the same the regression studies on each separately. The public sleeve is the portfolio of publicly traded stocks held inside Berkshire’s insurance subsidiaries — disclosed quarterly in SEC Form 13F filings. This is what most financial press coverage focuses on. The Coca-Cola, the American Express, the Apple, the Bank of America. Over the full sample it averaged about 35% of Berkshire’s total capital. The private sleeve is the portfolio of wholly-owned operating businesses — See’s Candies, Nebraska Furniture Mart, GEICO after the 1995 full acquisition, BNSF after 2010, Berkshire Hathaway Energy, Dairy Queen, NetJets, Precision Castparts, the whole roster of consolidated subsidiaries. Over the full sample the private portion grew from under 20% of Berkshire to more than 78% today. Berkshire was once an insurance company with an equity portfolio. Today it’s an insurance company owned by a conglomerate. And here's why that matters. The public sleeve — the portfolio of stocks Buffett bought fractionally and held — earned an average excess return of 12.0% per year at 16.2% volatility. Sharpe ratio: 0.74. The private sleeve — the portfolio of whole companies Buffett bought outright — earned an average excess return of 9.3% per year at 20.6% volatility. Sharpe ratio: 0.45. The publicly traded pieces of companies Buffett owned delivered materially better returns, especially when compared against the risk taken. The private sleeve’s Sharpe ratio of 0.45 is, remarkably, lower (worse) than the broad market’s 0.49 over the same period. In other words, when Buffett bought pieces of great public companies, he outperformed. When Buffett bought whole private companies, he did not. The private sleeve’s drag on Berkshire’s overall performance is meaningful. That drag was smaller in the early years, when the private portfolio was only 20% of the business. As the private sleeve has grown to 78% of Berkshire’s capital, the drag has grown proportionally. The declining Sharpe ratio of Berkshire over time — which every long-term shareholder has felt, even if they could not name it — comes primarily from the growing share of capital trapped inside whole-company acquisitions that underperform the public-market alternatives Buffett could have bought instead. Learn more about Warren's Mistakes and how to learn these lessons to improve your own investing in my new book.